Remember the lag in the polling data. Check out the LATimes flawed poll. The one value it has, to me, is that regardless that they may have a flawed pool, it's changing by the day.
Remember that the Dornsife (LATimes) poll is always polling the same 3000-people group, so they are a bunch who has an exaggerated awareness of what's going on (that's why there's a rule in the trade that says you shouldn't poll the same person more than three times in a row).
The general public is not that committed, but this poll, as you say, may give some hints of where is it the general mood going.
Additionally, besides their election forecast (to date, 45.3% Trump, 42.3% Clinton, with a record high 2711/3000 respondents) they have the poll about intention to vote by candidate (to date, 82.3% Clinton, 81% Trump) where you usually find that Trump supporters are not very enthusiastic about going and vote (not today).
I had this poll as skewed towards Trump by 3% when compared with the average of other polls I selected as reliable. To my satisfaction I saw one, I don't remember which one, of the many sites that collect polls (Huffington, Real Clear Politics, etc) making the exact same correction before entering it into their forecast.
But this Dornsife poll doesn't inform the content of their questionnaire. We don't know how they include third party candidates. That's very important. There are at least three different ways to ask the question in all those polls with dozens of variations: In one extreme there's one poll that asks "Clinton or Trump?" and doesn't accept any other option (a forced poll). They are giving Clinton 54%, Trump 46% these days. On the other extreme there are polls that list five candidates, give the options of not voting or not decided yet and ask as if the election were today. They are rendering Clinton 40%, Trump 39%, Johnson 8%, Stein 3%, Castle and others 1%, not decided 4% and not going to vote 5%. Reality is today somewhere between them.